Pyridoxinamine 5'-phosphate oxidases (P(N/M)P oxidases) that bind flavin mononucleotide (FMN) and oxidize pyridoxine 5'-phosphate or pyridoxamine 5'-phosphate to form pyridoxal 5'-phosphate (PLP) are an important class of enzymes that play a central role in cell metabolism. Failure to generate an adequate supply of PLP is very detrimental to most organisms and is often clinically manifested as a neurological disorder in mammals. In this study, we analyzed the function of YLR456W and YPR172W, two homologous genes of unknown function from S. cerevisiae that have been annotated as putative P(N/M)P oxidases based on sequence homology. Different experimental approaches indicated that neither protein catalyzes PLP formation nor binds FMN. On the other hand, our analysis confirmed the enzymatic activity of Pdx3, the S. cerevisiae protein previously implicated in PLP biosynthesis by genetic and structural characterization. After a careful sequence analysis comparing the putative and confirmed P(N/M)P oxidases, we found that the protein domain (PF01243) that led to the YLR456W and YPR172W annotation is a poor indicator of P(N/M)P oxidase activity. We suggest that a combination of two Pfam domains (PF01243 and PF10590) present in Pdx3 and other confirmed P(N/M)P oxidases would be a stronger predictor of this molecular function. This work exemplifies the importance of experimental validation to rectify genome annotation and proposes a revision in the annotation of at least 400 sequences from a wide variety of fungal species that are homologous to YLR456W and are currently misrepresented as putative P(N/M)P oxidases.